[SPIR-V] Add SPV_EXT_descriptor_heap + SPV_KHR_untyped_pointers codegen by jzakharovnv · Pull Request #8517 · microsoft/DirectXShaderCompiler

Jonathan Zakharov (jzakharovnv) · 2026-06-04T21:43:25Z

Building off of #8281, this PR adds a native lowering via SPV_EXT_descriptor_heap and SPV_KHR_untyped_pointers and is part 1/4 in a series.

ResourceDescriptorHeap and SamplerDescriptorHeap are lowered to untyped variables decorated with ResourceHeapEXT and SamplerHeapEXT. Each heap access emits OpUntypedAccessChainKHR into a runtime array of the appropriate descriptor type. Buffer-like resources (StructuredBuffer, ByteAddressBuffer, ConstantBuffer, TextureBuffer) use OpTypeBufferEXT and OpBufferPointerEXT; image and sampler resources use OpLoad. Interlocked operations on RWTexture use OpUntypedImageTexelPointerEXT.

Requires -fspv-target-env=vulkan1.3.

Assisted by an AI agent.

Diego Novillo (@dnovillo)

github-actions · 2026-06-04T21:45:20Z

✅ With the latest revision this PR passed the C/C++ code formatter.

Jonathan Zakharov (jzakharovnv) · 2026-06-04T23:02:24Z

@microsoft-github-policy-service agree company="NVIDIA"

Diego Novillo (dnovillo)

Thanks for this! I just started looking at it and have a couple of questions. I'll add more as I read the PRs.

Diego Novillo (dnovillo) · 2026-06-08T21:05:41Z

+SpirvEmitter::getDescriptorHeapRuntimeArrayType(const SpirvType *elemType,
+                                                bool onSamplerHeap) {
+  constexpr uint32_t kDefaultResourceHeapStride = 64;
+  constexpr uint32_t kDefaultSamplerHeapStride = 32;


I think I would float these defaults to SpirvEmitter.h and document where the seemingly magic values 32 and 64 come from.

Definitely agree that these need some better documentation. I placed them at this scope because they seem a little niche, and I want to try to avoid polluting very wide scopes, but I can move them up.

With the inclusion of OpConstantSizeOfEXT, per Tobski, these default stride constants have been removed.

Diego Novillo (dnovillo) · 2026-06-08T21:54:36Z

+  constexpr uint32_t kDefaultSamplerHeapStride = 32;
+  const uint32_t stride =
+      onSamplerHeap ? kDefaultSamplerHeapStride : kDefaultResourceHeapStride;
+  return spvContext.getRuntimeArrayType(elemType, stride);


It doesn't seem that there are tests for OpDecorate ArrayStride N in this PR. I haven't checked the others. Could you add one (unless I missed it?)

Yes, there is a healthy amount of testing in the last 2 PRs of this series, but I neglected to have for this one. I can fix that.

Chris B (llvm-beanz) · 2026-06-23T15:28:07Z

-    addExtension(Extension::EXT_descriptor_heap, "DescriptorHeap", {});
-    addExtension(Extension::KHR_untyped_pointers, "DescriptorHeap", {});
+    const llvm::StringRef feature = "DescriptorHeap";
+    featureManager.requestTargetEnv(SPV_ENV_VULKAN_1_3, feature, {});


This function can return failure (which is getting dropped here), and there is no test verifying its error handling.

Thanks for catching that, I'll add a test.

Handled with Guard requestTargetEnv return value for descriptor heap

Chris B (llvm-beanz) · 2026-06-23T15:32:16Z

+            tryToAssignToDescriptorHeapBuffer(expr))
+      return aliasResult.getValue();
+
+    auto *rhs = loadIfGLValue(expr->getRHS());


LLVM's coding standards which DXC adopts (although not historically well enforced), have an "almost never auto" policy:

https://github.com/microsoft/DirectXShaderCompiler/blob/main/docs/CodingStandards.rst#use-auto-type-deduction-to-make-code-more-readable

The heavy use of auto in this code makes it significantly harder to review this PR in a browser without an IDE telling me the types of things.

Noted, thanks for tip!

Handled with Replace auto with explicit types in descriptor heap code

The re-base removed the auto suggestion from Chris B (@llvm-beanz) Could you re-apply it?

Chris B (llvm-beanz) · 2026-06-23T15:33:30Z

+  if (result && !result->isRValue()) {
+    result =
+        spvBuilder.createLoad(structType, result, buffer->getExprLoc(), range);
+  }


Suggested change

if (result && !result->isRValue()) {

result =

spvBuilder.createLoad(structType, result, buffer->getExprLoc(), range);

}

if (result && !result->isRValue())

result =

spvBuilder.createLoad(structType, result, buffer->getExprLoc(), range);

nit: https://llvm.org/docs/CodingStandards.html#don-t-use-braces-on-simple-single-statement-bodies-of-if-else-loop-statements

Will fix in next commit

Handled with Remove unnecessary braces in descriptor heap code

Chris B (llvm-beanz) · 2026-06-23T15:35:31Z

+  }
+
+  // ConstantBuffer -> Uniform (UBO); all others -> StorageBuffer (SSBO)
+  // TODO: Remove this manual override once LowerTypeVisitor returns the


Should this be fixed before we merge this change? Seems like you're working around a clear bug here.

To get the storage class here rather than in LowerTypeVisitor is bit of a bandage but provides functionally correct output. I had left this TODO in with the aim of making this PR as small/unintrusive as possible and with room to improve with future commits. If you're okay with filing an issue, let me know, otherwise I can go about cleaning up this fix.

I think I'm okay with using an issue to track that as a follow up. Normally it would be better to do that foundational work first to avoid committing code in an undesirable state unless there's a reason why it can't be done before this change merged.

Understandable. I am open to clean the code up, it just may be a while before I am able to address everything (on the order of 2 weeks). Let me know if that is what you prefer!

Circling back! It has been a busy few weeks, but I am open to fixing this now if it means it could be reviewed before SIGGRAPH (July 19th). Would this be possible for you, or can we file an issue instead?

Handled with Fixed correct StorageClass for descriptor heap buffer alias pointers TODO

Now we have these fixes in #8519 but not here. This will give you merge conflicts. Are you thinking of submitting each of these PRs separately? If so, this is probably the natural first one to submit. In which case, you'll probably need to move the code from #8519 here.

Chris B (llvm-beanz) · 2026-06-23T15:41:19Z

+
+float4 main(uint idx : A) : SV_Target {
+  Texture2D<float4> tex = ResourceDescriptorHeap[NonUniformResourceIndex(idx)];
+  SamplerState samp = SamplerDescriptorHeap[NonUniformResourceIndex(idx + 1)];


This seems like something we should have the compiler issue a diagnostic on. Silently dropping something the user explicitly wrote seems unfortunate.

I agree, this is a bit esoteric but correct according to Tobski. How should we handle a diagnostic here, just a simple warning?

Oh... Actually I think the correct solution for HLSL is that in the absence of NonUniformResourceIndex an access should be marked as Uniform, and NonUniformResourceIndex suppresses that.

It should have an effect, so no diagnostic should be required.

Tobski · 2026-07-07T16:09:15Z

Jonathan Zakharov (@jzakharovnv) I think the OpConstantSizeOfEXT usage in this MR isn't quite right - it looks like you're just taking the stride of the descriptor type being loaded and using that? This works for the sampler heap where there's only one type (samplers), but not for the resource heap.

The stride of all resource descriptors must be the same for all descriptors, and has to be based on the biggest of the buffer/image sizes. This is illustrated in the VK extension doc here: https://docs.vulkan.org/features/latest/features/proposals/VK_EXT_descriptor_heap.html#_shader_model_6_6_samplerheap_and_resourceheap

I might be misunderstanding the code, but I don't see anything for finding the largest of the two sizes? They're both POT, so it is a simple "which is max" calculation, but I don't see it being done in this PR.

Jonathan Zakharov (jzakharovnv) · 2026-07-07T17:43:11Z

Tobski You are right, this is my blunder. Will fix shortly.

Jonathan Zakharov (jzakharovnv) · 2026-07-10T22:35:26Z

Tobski Shared max(image,buffer) stride for descriptor heap resource arrays ought to be closer to the original intent. Please take a look when you can. Thanks!

Tobski · 2026-07-13T11:15:43Z

Tobski Shared max(image,buffer) stride for descriptor heap resource arrays ought to be closer to the original intent. Please take a look when you can. Thanks!

Looks right to me now! Thanks!

Diego Novillo (dnovillo)

A question on changes that have gone in #8519 that we may to reflect here. Not sure how you want to handle it.

Diego Novillo (dnovillo) · 2026-07-17T16:48:26Z

+  }
+
+  // ConstantBuffer -> Uniform (UBO); all others -> StorageBuffer (SSBO)
+  // TODO: Remove this manual override once LowerTypeVisitor returns the


Now we have these fixes in #8519 but not here. This will give you merge conflicts. Are you thinking of submitting each of these PRs separately? If so, this is probably the natural first one to submit. In which case, you'll probably need to move the code from #8519 here.

Jonathan Zakharov (jzakharovnv) · 2026-07-20T21:48:54Z

Diego Novillo (@dnovillo) In response to your last comment on this PR, yes I think it's more correct to have the TODO commit a part of this initial branch rather than one down the line. Cherrypicked and rebased to reflect this.

Diego Novillo (dnovillo)

Just one minor change and it's good to from my side.

Diego Novillo (dnovillo) · 2026-07-21T20:08:13Z

+            tryToAssignToDescriptorHeapBuffer(expr))
+      return aliasResult.getValue();
+
+    auto *rhs = loadIfGLValue(expr->getRHS());


The re-base removed the auto suggestion from Chris B (@llvm-beanz) Could you re-apply it?

Diego Novillo (dnovillo)

LGTM. Just one final nit. Thanks for doing this!

Diego Novillo (dnovillo) · 2026-07-24T12:21:30Z

+  if (result && !result->isRValue()) {
+    result =
+        spvBuilder.createLoad(structType, result, buffer->getExprLoc(), range);
+  }


Could you re-apply the suggestion from Chris B (@llvm-beanz)? It got dropped.

Building off of microsoft#8281, this commit adds a native lowering via SPV_EXT_descriptor_heap and SPV_KHR_untyped_pointers. ResourceDescriptorHeap and SamplerDescriptorHeap are lowered to untyped variables decorated with ResourceHeapEXT and SamplerHeapEXT. Each heap access emits OpUntypedAccessChainKHR into a runtime array of the appropriate descriptor type. Buffer-like resources (StructuredBuffer, ByteAddressBuffer, ConstantBuffer, TextureBuffer) use OpTypeBufferEXT and OpBufferPointerEXT; image and sampler resources use OpLoad. Interlocked operations on RWTexture use OpUntypedImageTexelPointerEXT. Requires -fspv-target-env=vulkan1.3. Assisted-by: Claude.

…TODO

Damyan Pepper (damyanp) · 2026-07-30T16:48:15Z

[Auto-generated note from Damyan Pepper (@damyanp)]

This looks like a user-visible bug fix/feature change. Please add (or point to) the corresponding entry in docs/ReleaseNotes.md.

If release-note coverage is planned in a related PR (including one that hasn’t been submitted yet), please mention that plan/link so we can avoid duplicate notes.

Copilot

Pull request overview

Adds native SPIR-V descriptor-heap lowering using SPV_EXT_descriptor_heap and SPV_KHR_untyped_pointers.

Changes:

Implements native image, sampler, and buffer descriptor access.
Adds descriptor-size-based runtime-array strides and image atomic support.
Expands tests and documents Vulkan 1.3 requirements and limitations.

Reviewed changes

Copilot reviewed 47 out of 47 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
`tools/clang/unittests/SPIRV/SpirvContextTest.cpp`	Tests runtime-array type uniqueness.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.typed-formats.hlsl`	Tests typed image formats.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.texturecube.hlsl`	Tests cube textures and samplers.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.texture.hlsl`	Tests texel buffers.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.texture-sampler-assignment.hlsl`	Tests resource reassignment.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.texture-ms.hlsl`	Tests multisampled textures.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.texture-dims.hlsl`	Tests sampled texture dimensions.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.structured-buffer-atomic.hlsl`	Tests structured-buffer atomics.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.static-global.hlsl`	Tests static global resources.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.sampler-comparison.hlsl`	Tests comparison samplers.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.sample-grad-bias.hlsl`	Tests sampling operands.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.rwtexture-dims.hlsl`	Tests RW texture dimensions.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.rwtexture-atomics.hlsl`	Tests untyped image atomics.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.rwbyteaddressbuffer.hlsl`	Tests writable byte buffers.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.nonuniform.hlsl`	Tests divergent heap indexing.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.mixed-bound.hlsl`	Tests bound/native resource coexistence.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.mixed-alias.error.hlsl`	Tests mixed-alias diagnostics.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.load-offset.hlsl`	Tests texture load offsets.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.groupshared.hlsl`	Tests groupshared coexistence.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.gather.hlsl`	Tests gather operations.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.function-params.hlsl`	Tests resource function parameters.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.discarded.error.hlsl`	Tests discarded-index diagnostics.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.counter-ops.error.hlsl`	Tests unsupported counter diagnostics.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.constant-texture-buffer.hlsl`	Tests constant and texture buffers.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.constant-buffer-assignment.hlsl`	Tests constant-buffer reassignment.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.buffer.hlsl`	Tests native buffer lowering.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.array-stride.hlsl`	Tests descriptor-array strides.
`tools/clang/test/CodeGenSPIRV/sm6_6.descriptorheap.ext.append-consume.error.hlsl`	Tests append/consume diagnostics.
`tools/clang/test/CodeGenSPIRV/resource-heap-ext-texture.hlsl`	Removes superseded coverage.
`tools/clang/lib/SPIRV/SpirvType.cpp`	Extends runtime-array equality.
`tools/clang/lib/SPIRV/SpirvInstruction.cpp`	Implements new SPIR-V instructions.
`tools/clang/lib/SPIRV/SpirvEmitter.h`	Declares heap lowering and alias state.
`tools/clang/lib/SPIRV/SpirvEmitter.cpp`	Implements native heap code generation.
`tools/clang/lib/SPIRV/SpirvContext.cpp`	Uniques buffer and stride-ID types.
`tools/clang/lib/SPIRV/SpirvBuilder.cpp`	Builds descriptor sizes and strides.
`tools/clang/lib/SPIRV/LowerTypeVisitor.cpp`	Lowers untyped image pointers.
`tools/clang/lib/SPIRV/EmitVisitor.h`	Declares new emission handlers.
`tools/clang/lib/SPIRV/EmitVisitor.cpp`	Serializes new instructions and decorations.
`tools/clang/lib/SPIRV/DeclResultIdMapper.h`	Declares function alias registration.
`tools/clang/lib/SPIRV/DeclResultIdMapper.cpp`	Implements alias registration.
`tools/clang/lib/SPIRV/CapabilityVisitor.cpp`	Requires Vulkan 1.3 and extensions.
`tools/clang/include/clang/SPIRV/SpirvVisitor.h`	Adds visitor hooks.
`tools/clang/include/clang/SPIRV/SpirvType.h`	Adds stride-ID runtime arrays.
`tools/clang/include/clang/SPIRV/SpirvInstruction.h`	Defines new instruction classes.
`tools/clang/include/clang/SPIRV/SpirvContext.h`	Adds type caches and APIs.
`tools/clang/include/clang/SPIRV/SpirvBuilder.h`	Exposes descriptor-stride builders.
`docs/SPIR-V.rst`	Documents native descriptor heaps.

+  const bool srcIsHeap = isDescriptorHeap(srcExpr->IgnoreParenCasts());
+  const bool wasHeap = descriptorHeapImageAliasVars.count(dstVar) ||
+                       descriptorHeapBufferAliasVars.count(dstVar);
+  const bool wasBound = stateIt != descriptorHeapVarState.end() &&
+                        stateIt->second == DescriptorHeapVarState::Bound;
+  const bool mixingDetected =
+      (srcIsHeap && wasBound) || (!srcIsHeap && wasHeap);


+    } else if (const auto *varDecl = dyn_cast<VarDecl>(decl)) {
+      if (auto *alias =
+              emitDescriptorHeapBufferPointer(varDecl, expr->getLocStart()))
+        result = alias;


        const SpirvType *handleType =
            lowerTypeVisitor.lowerType(resourceType, SpirvLayoutRule::Void,
                                       llvm::None, baseExpr->getExprLoc());
-        const auto *arrayType =
-            spvContext.getRuntimeArrayType(handleType, llvm::None);
-        auto *untypedAccessChainPtr = spvBuilder.createUntypedAccessChainKHR(
-            untypedType, arrayType, var, index, baseExpr->getExprLoc());
+        const SpirvType *arrayType =
+            getDescriptorHeapRuntimeArrayType(handleType);


+A local resource variable initialized from a heap access is resolved entirely at
+compile time: the variable is recorded as an alias for the heap index, and every
+later use is re-lowered as a fresh access chain rather than as a load of a stored
+descriptor handle. This is sound only when the variable holds a heap descriptor


+// CHECK-DAG:       %[[SampSize:[a-zA-Z0-9_]+]] = OpConstantSizeOfEXT %uint %[[SamplerDesc]]
+
+// Every resource runtime array shares the one resource stride. 
+// The sampler array just uses the sampler size as it's stride.


Fixes #8603: MergeBinaryOpSelect produced an OpSelect with a vector result type and a scalar condition, which is only legal in SPIR-V 1.4+ (KhronosGroup/SPIRV-Tools#6827). This exposes #8740 where the resource-heap-ext-texture.hlsl test fails validation - for now this is worked around by disabling the test, since #8517 looks like it is reworking it anyway. Assisted-by: copilot --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: b04b21af-2d8b-4af5-b9ad-1b84073588f5

Diego Novillo (dnovillo) · 2026-08-06T22:36:43Z

Jonathan Zakharov (@jzakharovnv) Damyan Pepper (@damyanp) Gregory Roth (@pow2clk), I have gone over all the open review threads for all the PRs in this series #8517, #8518 and #8519. I would like to get them to the point where we can submit them so that other developers can start testing this implementation. I went over the reviews and tried to separate blocking issues from issues we could postpone in a follow-up PR or even open an issue to fix later.

I would like to get consensus on what we need done to get some forward progress. Given that this feature is protected by a flag, I don't think is should be too destabilizing to add it even when there are still things to fix. I don't want to have a broken initial version, but I also don't want to spend a long time getting it to be perfect before it goes in. We need the feature exposed to developers so they can start testing it.

I'll add all my notes here and will put cross-references in the other PRs so I don't lose track. These are the things I think should be fixed now before it goes in. These were signaled by Copilot on 4/Aug:

Two blocking issues for #8517

The srcIsHeap misclassification. This one is a false error on valid HLSL, and would affect resource kinds that use aliases. May be do the check based on heap provenance rather than on the source being a subscript expression?
The unregistered-decl path, which you already said you would fix. We don't need full support now. A clean diagnostic and a not %dxc test is enough, with an issue for the real fix.

One blocking issue for #8518

The unregistered-decl path. This one could be converted into a diagnostic with a not %dxc test for now. The same defect shows up on [SPIR-V] Add SPV_EXT_descriptor_heap + SPV_KHR_untyped_pointers codegen #8517, so wherever you fix it is fine by me. An issue to follow-up would be enough for now.

Two blocking issues for #8519

The acceleration-structure diagnostic firing under -fvk-resource-heap-stride. We are now rejecting valid use of this feature. Should be a straightforward fix adding && !spirvOptions.resourceHeapStride.has_value(), plus a RUN line.
The release note. CONTRIBUTING.md. does call for one on new compiler options, so I would just add it.

Everything else I am happy to defer, and I do not think it should hold back the series:

Chris B (@llvm-beanz)'s requestTargetEnv point and brace suggestion. You fixed both in 6bc5a9e and a later rebase dropped them, along with sm6_6.descriptorheap.ext.targetenv.error.hlsl. Could you fold them back in next time you touch the branch, and check whether the rebases lost anything else?
The NonUniformResourceIndex question. This needs more discussion. The current behaviour matches what Tobski described and is documented. One thing we could have an issue for: the index OpCopyObject still carries NonUniformEXT, so modules pick up ShaderNonUniformEXT and SPV_EXT_descriptor_indexing that they do not need.
Acceleration structures missing from the stride. [SPIR-V] Add descriptor heap RaytracingAccelerationStructure support #8518 fixes this, so it is only a problem if [SPIR-V] Add SPV_EXT_descriptor_heap + SPV_KHR_untyped_pointers codegen #8517 is submitted on its own.
The SPIR-V.rst wording and the it's typo. Cosmetic, follow-up whenever.
The acceleration structure reassignment rejection. The invalid SPIR-V is fixed, so we just have a misleading message. A follow-up PR can fix this.
The parenthesized subscript rejection. This one predates these PRs, so it is not a regression here.
The run-together error messages. Only reachable on an already-invalid command line, and both messages are still readable.
The missing one-flag RUN lines. The fallback seems right. This seems more of a completeness thing than a risk.
The untested requires -spirv branch. It is a copy of two existing branches in that file, neither of which is tested either.
Two more that Copilot raised and you have already tracked as issues: the out/inout writeback and uninitialized-declaration cases in #8732, and the stride being derived from compiler flags rather than the heap layout in #8714. Thanks for filing those.

Five issues triaged in isolated parallel sessions against a Debug build of ab54009, collated by a separate session briefed only by on-disk artifacts. | # | verdict | history | | --- | --- | --- | | 2530 | repros | always (20/20 releases) | | 3055 | repros | v1.4.1907 output byte-identical to main | | 3259 | repros | always from v1.5.2010 | | 8725 | repros | always from v1.8.2505 (5 of 20 can express lib_6_9) | | 8732 | inconclusive | unmeasurable -- filed against unmerged PR microsoft#8517 | microsoft#8732 is the useful outlier: every symbol it blames has zero occurrences on main, so it is not does-not-repro -- the symptom was never in the compiler being measured. It needs a human decision, not a fix. The headline tooling fix is the invalid-probe classifier, which batch 004 predicted would be least trustworthy where the reported symptom is itself a diagnostic. It was, in both directions: a release emitting a correct, complete diagnostic -- what a fix looks like on such an issue -- scored invalid-probe and could hide the fixing release; and a probe whose predicate had *matched* was still discarded because the symptom text is itself a marker, erasing ground truth and then misattributing it to the profile. microsoft#3055 escaped only because dxc says "no matching member function for call to" where the marker is "no matching function for call to". The fix is deliberately narrow: a demotion is suppressed only when the issue's own match.json positively quotes the marker text. Making the classifier broadly more permissive would reintroduce the fake-regression bug it exists to prevent, which has already produced wrong verdicts twice. Verified against every archived predicate -- no legitimate demotion is lost, and a reindex moved exactly the two probes that demonstrated the defect out of 346. Also fixed: reindex now re-scores variant captures, which previously were never checked against the current classifier -- its first run found three stale headers in microsoft#2202 that had been wrong since batch 004; invalid-probe verdicts now record why on disk; `is not supported` is anchored to a target/profile clause; ce_args() valueless-flag handling; a warning when --args diverges from cmd.txt; a labels --refresh REST fallback. ~30 new tests. Ground truth was rebuilt for this batch (upstream/main merged, SPIRV-Tools submodule to 1c336172, cached version headers cleared) because microsoft#8732 sits in the blast radius of ec2ba18. That proved load-bearing rather than precautionary: the reporter's own workaround compiles on v1.9.2607 but fails on main under a newly enforced UniformConstant ArrayStride rule. The stale build would have reported the feature's state incorrectly. Batch 005's verdicts are therefore measured against a different compiler than batches 001-004; triaged_with_commit records this per issue. Report-only: no issue was edited, commented on or closed, and no DXC source is touched. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6bd9cb60-aeb2-4d6c-9462-7a5e79fe1cda

github-project-automation Bot added this to HLSL Roadmap Jun 4, 2026

github-project-automation Bot moved this to New in HLSL Roadmap Jun 4, 2026

Jonathan Zakharov (jzakharovnv) mentioned this pull request Jun 4, 2026

[SPIR-V] Add descriptor heap RaytracingAccelerationStructure support #8518

Open

Jonathan Zakharov (jzakharovnv) force-pushed the pr1-descriptor-heap-core branch 2 times, most recently from f640327 to cbcae38 Compare June 4, 2026 22:42

Diego Novillo (dnovillo) suggested changes Jun 8, 2026

View reviewed changes

github-project-automation Bot moved this from New to In progress in HLSL Roadmap Jun 8, 2026

Chris B (llvm-beanz) reviewed Jun 23, 2026

View reviewed changes

Jonathan Zakharov (jzakharovnv) mentioned this pull request Jun 26, 2026

[SPIR-V] Add descriptor heap stride specialization constant attribute #8520

Closed

Diego Novillo (dnovillo) suggested changes Jul 17, 2026

View reviewed changes

Diego Novillo (dnovillo) added the spirv Work related to SPIR-V label Jul 17, 2026

Jonathan Zakharov (jzakharovnv) force-pushed the pr1-descriptor-heap-core branch from bd79339 to cbcae38 Compare July 17, 2026 20:35

Diego Novillo (dnovillo) suggested changes Jul 21, 2026

View reviewed changes

Diego Novillo (dnovillo) approved these changes Jul 24, 2026

View reviewed changes

Jonathan Zakharov (jzakharovnv) added 6 commits July 28, 2026 13:51

Added OpConstantSizeOfEXT

c7d2389

clang-format

153e928

Shared max(image,buffer) stride for descriptor heap resource arrays

eb4a9bc

Fixed correct StorageClass for descriptor heap buffer alias pointers …

25c5a0a

…TODO

Ammend keyword usage to follow LLVM Coding Standards

88bd33c

Jonathan Zakharov (jzakharovnv) force-pushed the pr1-descriptor-heap-core branch from d0835c8 to 88bd33c Compare July 28, 2026 21:00

fix stride test

1007865

Jonathan Zakharov (jzakharovnv) added 2 commits August 3, 2026 11:07

Cover arrayStrideId uniquing in SpirvContextTest

f502bcf

Diagnose mixed bound/heap aliasing and correct documentation

034f3c5

Copilot AI review requested due to automatic review settings August 4, 2026 00:41

Copilot started reviewing on behalf of Jonathan Zakharov (jzakharovnv) August 4, 2026 00:41 View session

Copilot AI reviewed Aug 4, 2026

View reviewed changes

Jonathan Zakharov (jzakharovnv) mentioned this pull request Aug 4, 2026

[SPIR-V] SPV_EXT_descriptor_heap mixed bound/heap aliasing causes silent miscompilation or ICE #8732

Open

This was referenced Aug 6, 2026

Update SPIRV-Tools to 1c336172 #8738

Merged

[SPIR-V] DXC SPIR-V test fail with the latest spir-v tools #8740

Open

Diego Novillo (dnovillo) mentioned this pull request Aug 6, 2026

[SPIR-V] Add descriptor heap -fvk-resource-heap-stride / -fvk-sampler-heap-stride CLI flags #8519

Open

Uh oh!

Conversation

Jonathan Zakharov (jzakharovnv) commented Jun 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jun 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Jonathan Zakharov (jzakharovnv) commented Jun 4, 2026

Uh oh!

Diego Novillo (dnovillo) left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Tobski commented Jul 7, 2026

Uh oh!

Jonathan Zakharov (jzakharovnv) commented Jul 7, 2026

Uh oh!

Jonathan Zakharov (jzakharovnv) commented Jul 10, 2026

Uh oh!

Tobski commented Jul 13, 2026

Uh oh!

Diego Novillo (dnovillo) left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Jonathan Zakharov (jzakharovnv) commented Jul 20, 2026

Uh oh!

Diego Novillo (dnovillo) left a comment

Choose a reason for hiding this comment

Jonathan Zakharov (jzakharovnv) commented Jun 4, 2026 •

edited

Loading

github-actions Bot commented Jun 4, 2026 •

edited

Loading